What does the "Reference to non-existent field 'training' " error message mean?

hi, can you everyone help me in this error.

the code is:

% Common parameter settings
opts.k  = 5;      % Number of k in K-nearest neighbor
opts.N  = 10;     % number of solutions
opts.T  = 100;    % maximum number of iterations
opts.P     = 0.5;    % constant
opts.FADs  = 0.2;    % fish aggregating devices effect
% Divide data into training and validation sets
HO = cvpartition(trainingLabels,'HoldOut',.2); 
opts.Model = HO; 
% Perform feature selection 
FS = jfs('mpa',trainingFeatures,trainingLabels,opts);
% Define index of selected features
sf_idx = FS.sf;
% Accuracy
Acc = jknn(feat(:,sf_idx),trainingLabels,opts); 

call function:

function model = jfs(type,feat,label,opts)
switch type
    case 'mpa'  ; fun = @jMarinePredatorsAlgorithm; 
        case 'gndo' ; fun = @jGeneralizedNormalDistributionOptimization;
            case 'sma'  ; fun = @jSlimeMouldAlgorithm; 
end
tic;
model = fun(feat,label,opts); 
% Computational time
t = toc;
model.t = t;
fprintf('\n Processing Time (s): %f % \n',t); fprintf('\n');
end

call function:

function MPA = jMarinePredatorsAlgorithm(feat,label,opts)
% Parameters
lb    = 0;
ub    = 1; 
thres = 0.5; 
beta  = 1.5;   % levy component
P     = 0.5;   % constant
FADs  = 0.2;   % fish aggregating devices effect
if isfield(opts,'N'), N = opts.N; end
if isfield(opts,'T'), max_Iter = opts.T; end
if isfield(opts,'thres'), thres = opts.thres; end
if isfield(opts,'P'), P = opts.P; end
if isfield(opts,'FADs'), FADs = opts.FADs; end

% Objective function
fun = @jFitnessFunction; 
% Number of dimensions
dim = size(feat,2); 
% Initial (9)
X   = zeros(N,dim);  
for i = 1:N
	for d = 1:dim
    X(i,d) = lb + (ub - lb) * rand(); 
  end
end
% Pre
fit  = zeros(1,N);
fitG = inf;

curve = inf; 
t = 1; 
% Iteration
while t <= max_Iter
  % Fitness
  for i=1:N
    fit(i) = fun(feat,label,(X(i,:) > thres),opts);
    % Best
    if fit(i) < fitG
      fitG = fit(i); 
      Xgb  = X(i,:);
    end
  end
  % Memory saving
  if t == 1
    fitM = fit; 
    Xmb  = X;
  end
  for i = 1:N
    if fitM(i) < fit(i)
      fit(i) = fitM(i);
      X(i,:) = Xmb(i,:);
    end
  end
  Xmb  = X;
  fitM = fit;
  % Construct elite (10)
  Xe   = repmat(Xgb,[N 1]);
  % Adaptive parameter (14)
  CF   = (1 - (t / max_Iter)) ^ (2 * (t / max_Iter));
  % [First phase] (12)
  if t <= max_Iter / 3
    for i = 1:N
      % Brownian random number
      RB = randn(1,dim);  
      for d = 1:dim
        R        = rand();
        stepsize = RB(d) * (Xe(i,d) - RB(d) * X(i,d));
        X(i,d)   = X(i,d) + P * R * stepsize;
      end
      % Boundary
      XB = X(i,:); XB(XB > ub) = ub; XB(XB < lb) = lb; 
      X(i,:) = XB;
    end
  % [Second phase] (13-14)
  elseif t > max_Iter / 3 && t <= 2 * max_Iter / 3
    for i = 1:N
      % First half update (13)
      if i <= N / 2
        % Levy random number
        RL = 0.05 * jLevy(beta,dim);
        for d = 1:dim
          R        = rand();
          stepsize = RL(d) * (Xe(i,d) - RL(d) * X(i,d));
          X(i,d)   = X(i,d) + P * R * stepsize;
        end
      % Another half update (14)
      else
        % Brownian random number
        RB = randn(1,dim); 
        for d = 1:dim
          stepsize = RB(d) * (RB(d) * Xe(i,d) - X(i,d));
          X(i,d)   = Xe(i,d) + P * CF * stepsize;
        end
      end
      % Boundary
      XB = X(i,:); XB(XB > ub) = ub; XB(XB < lb) = lb; 
      X(i,:) = XB;
    end
  % [Third phase] (15)
  elseif t > 2 * max_Iter / 3
    for i = 1:N
      % Levy random number
      RL = 0.05 * jLevy(beta,dim);
      for d = 1:dim
        stepsize = RL(d) * (RL(d) * Xe(i,d) - X(i,d));
        X(i,d)   = Xe(i,d) + P * CF * stepsize;
      end
      % Boundary
      XB = X(i,:); XB(XB > ub) = ub; XB(XB < lb) = lb; 
      X(i,:) = XB;
    end
  end 
  % Fitness
  for i = 1:N
    fit(i) = fun(feat,label,(X(i,:) > thres),opts);
    % Best
    if fit(i) < fitG
      fitG = fit(i);
      Xgb  = X(i,:);
    end
  end
  % Memory saving
  for i = 1:N
    if fitM(i) < fit(i)
      fit(i) = fitM(i); 
      X(i,:) = Xmb(i,:);
    end
  end
  Xmb  = X;
  fitM = fit;
  % Eddy formation and FADs effect (16)
  if rand() <= FADs
    for i = 1:N
      % Compute U
      U = rand(1,dim) < FADs;
      for d = 1:dim
        R      = rand();
        X(i,d) = X(i,d) + CF * (lb + R * (ub - lb)) * U(d);
      end
      % Boundary
      XB = X(i,:); XB(XB > ub) = ub; XB(XB < lb) = lb; 
      X(i,:) = XB;
    end
  else
    % Uniform random number [0,1]
    r   = rand(); 
    % Define two prey randomly
    Xr1 = X(randperm(N),:);
    Xr2 = X(randperm(N),:);
    for i = 1:N   
      for d = 1:dim
        X(i,d) = X(i,d) + (FADs * (1 - r) + r ) * ...
          (Xr1(i,d) - Xr2(i,d));
      end
      % Boundary
      XB = X(i,:); XB(XB > ub) = ub; XB(XB < lb) = lb;
      X(i,:) = XB;
    end
  end
  % Save
  curve(t) = fitG;
  fprintf('\nIteration %d Best (MPA)= %f',t,curve(t))
  t = t + 1;
end
% Select features based on selected index
Pos   = 1:dim;
Sf    = Pos((Xgb > thres) == 1); 
sFeat = feat(:,Sf); 
% Store results
MPA.sf = Sf; 
MPA.ff = sFeat; 
MPA.nf = length(Sf); 
MPA.c  = curve; 
MPA.f  = feat;
MPA.l  = label;
end
% Levy distribution
function LF = jLevy(beta,dim)
num   = gamma(1 + beta) * sin(pi * beta / 2); 
deno  = gamma((1 + beta) / 2) * beta * 2 ^ ((beta - 1) / 2); 
sigma = (num / deno) ^ (1 / beta);
u     = random('Normal',0,sigma,1,dim);
v     = random('Normal',0,1,1,dim);
LF    = u ./ (abs(v) .^ (1 / beta));
end

[Walter Roberson]

what shows up for

which -all cvpartition

[lebed toufik]

@Walter Roberson Show us about cvpartition

[Image Analyst]

@lebed toufik as you can see, HO does not have a field called training. Not sure why without digging into the code a lot more, which I don't have time to do now.

[lebed toufik]

@Image Analyst thank you so much.

[Walter Roberson]

At least in current versions, training is a method for the kinds of objects returned by cvpartition, so HO.training would be equivalent to HO.training() which would be equivalent to training(HO)

If @lebed toufik had shown us the result of

which -all cvpartition

as I had asked, then we would have been able to verify that the correct version of cvpartition was being called, and we would have been able to see the MATLAB version being used (since the user did not fill in the release information in the question.)

[lebed toufik]

@Walter Roberson the MATLAB version used is R2016a

[lebed toufik]

@Walter Roberson

[Image Analyst]

It really baffles and stuns me that you won't even answer @Walter Roberson's simple, very direct and very explicit question about what you see when you do this in the command window:

which -all cvpartition

Can you explain why you refuse to do that so he can help you?

[lebed toufik]

@Walter Roberson @Image Analyst sorry for the late reply, when i do this in the command window:

Answer 1

[Image Analyst]

Early in the program (line 8) you do this

HO = cvpartition(trainingLabels,'HoldOut',.2);

however you never defined/assigned trainingLabels, so it throws an error:

Unrecognized function or variable 'trainingLabels'.

Error in test7 (line 8)

HO = cvpartition(trainingLabels,'HoldOut',.2);

You need to assign trainingLabels to something.

Answer 2

[Image Analyst]

OK, so you seem to have the cvpartition.m file but it's not in the place where mine is. You seem to have the toolbox installed off the root of your E drive instead of under "C:\Program Files\MATLAB\R2022b\toolbox" like we do.

>> which -all cvpartition

C:\Program Files\MATLAB\R2022b\toolbox\stats\stats\cvpartition.m % cvpartition constructor

C:\Program Files\MATLAB\R2022b\toolbox\stats\bigdata\@tall\cvpartition.m % tall method

Do you know why that is?

What does this show for you?

[status, result] = system('set PATH');
executableFolder = char(regexpi(result, 'Path=(.*?);', 'tokens', 'once'))

I see

> [status, result] = system('set PATH');

executableFolder = char(regexpi(result, 'Path=(.*?);', 'tokens', 'once'))

executableFolder =

'C:\Program Files\MATLAB\R2022b\bin\win64'

And if you go to that 'C:\Program Files\MATLAB\R2022b' folder, do you see a toolbox subfolder there?

If you need immediate help, call the tech support telephone line.

[lebed toufik]

@Image Analyst

[Image Analyst]

It could be due to your using a 7 year old version, r2016a. I suggest you upgrade.

If you need immediate help, call the tech support telephone line.

[lebed toufik]

@Image Analyst thank you so much.